Re: [PATCH RFC net-next] tracing: accelerate tracing filters with BPF

[Alexei Starovoitov]

On Tue, May 13, 2014 at 10:09 PM, Ingo Molnar <mingo@xxxxxxxxxx> wrote:
>
> * Alexei Starovoitov <ast@xxxxxxxxxxxx> wrote:
>
>> On Tue, May 13, 2014 at 8:17 PM, Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
>> > On Tue, 13 May 2014 19:55:11 -0700
>> > Alexei Starovoitov <ast@xxxxxxxxxxxx> wrote:
>> >
>> >> Tracing filters are parsing user supplied character string and constructing
>> >> a predicate tree. filter_match_preds() was used to walk nodes of the tree to
>> >> simulate matching of boolean expression.
>> >>
>> >
>> > I applied this on top of net-next, from:
>> >
>> >  git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next.git
>> >  (master branch)
>> >
>> > and got this compile error:
>>
>> Sorry I wasn't clear in the log.
>> This patch depends on JIT patches that were posted separately.
>> Since this one is RFC, I didn't want to mix them.
>> JIT patches:
>> http://patchwork.ozlabs.org/patch/348558/
>> http://patchwork.ozlabs.org/patch/348559/
>> http://patchwork.ozlabs.org/patch/348560/
>
> Might make sense for you to get a korg account and stick such bits
> into Git trees, to ease testing.

it would be an honor. Thank you.

What is the next step for this particular patch?
Sounds like dependency on full net is a showstopper.
I think splitting bpf into small component will clean up
dependency for seccomp as well.
bpf can move into net/bpf/ dir and go via net-next tree.
If Dave is ok with this, I can start hacking.
It looks to be more or less straight forward.
There are few ways to split it. The only contentious piece
is skb_copy_bits(). I think we can come up with something.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/